3 ms·
It's what would in most industries be considered a reasonable sized sampling. When you then look at this poll then, things start to look grim indeed: http://www
by mladenkovacevic 13y ago
It's what would in most industries be considered a reasonable sized sampling. When you then look at this poll then, things start to look grim indeed: http://www.people-press.org/2013/06/10/majority-views-nsa-phone-tracking-as-acceptable-anti-terror-tactic/ http://www.people-press.org/2013/06/10/majority-views-nsa-ph...
- aagha 13y agoA reasonable sample is a function of the size of the overall population. Assuming you're talking a few million visitors in some fixed period of time, at 500 comments, you're still a little ways off from a 95 percent confidence interval and a +/-3 margin of error.
- caf 13y agoFor large populations it's not strongly a function of population size - a 500 sample is almost as good for a 100,000 population as it is for a 100,000,000 population (which is not very good at all).
- prostoalex 13y agoThis is a self-selected sample. People who read the article and thought "meh, whatever" did not bother to leave a comment.
- TeMPOraL 13y agoExactly. See also "voluntary response data are worthless" by prof. Velleman (in the tokenadult comment): https://news.ycombinator.com/item?id=2772533 https://news.ycombinator.com/item?id=2772533